[AusRace] Place Power - a system

Wed Apr 18 14:18:57 AEST 2018

Tony,

The apparent imbalance between %age of win pools and place pools is somewhat
a reflection of reality.

I first saw a "mathematical" calculation of one-race exotic odds based on
win odds in Scott's book - I don't know whether he devised the formulae, or
got them from another's work, but I instinctively knew they were flawed, and
my better calculations held me in good stead for many years. But I can't
beat 20% take-out by enough to bother.

To explain via a straightforward example of the simplest one-race exotic to
calculate - the Exacta.

Let's say we have calculated a 100% win market, (or SP adjusted to 100%):

1/1,  5/1,  10/1,  10/1,  20/1,  20/1,  20/1,  100/1 (=100.124298%)

Let's say the 100/1 shot wins

Scott then says you have a "race" between the other 7 for second, so
ignoring the 0.990099% reduction to 99.1432%, Scott says the 1/1 is still
1/1 and the 5/1 shot still 5/1 etc.

It just isn't so. 47% from my very large sample that start at 1/1 win (SP
market, not adjusted to 100%), whereas only 38% of those that do not win
come second, viz "win" the race for second, whereas it should be around the
same 47% if 100/1 shots won the races the 1/1 shot lost, even higher in
practise - eg in the above race, say the 5/1 shot wins, Scott takes the 17%
of the 5/1 shot from 100%, giving 83% and says the chances of the 1/1 shot
coming second having not won is 50/83 = 7/10, but they come second nothing
like that often.

I'm a mere statistician by training and practise, and seldom watch races,
let alone base my ratings on what I have seen, but my explanation is that
whatever prevented them from winning often also prevented them from coming
second (eg interference, missing the start, being poorly ridden, sea gulls,
being a stallion when a mare comes on heat..).

At the other end, the opposite applies - a 100/1 shot is 2.5 times as likely
to run second having lost, as it is to win, while a 10/1 chance is 1.43
times as likely to run second having lost as it is to win. 

LBL

From: Racing <racing-bounces at ausrace.com> On Behalf Of Tony Moffat
Sent: Tuesday, 17 April 2018 2:21 PM
To: racing at ausrace.com
Subject: Re: [AusRace] Place Power - a system

Len - thanks

I did not intend to discombobulate, flummox even - the system instruction is
to use those runners with the ratio of 1.2 or better and then I went with
1.12 or similar.

The idea was good, the execution not so.

No, I did not sit a shift at the keyboard watching these, I did watch a
couple of closing minutes for a few races, betting off my own ratings of
course

so I saw the trending for those I observed (to state the obvious) - if you
must know I spent most of the day sieving compost -  and listening to an
album by Hem.

(For broad beans Northerly, yes, I wore a mask.)

The 'Dr Z System' focuses on discrepancies between betting patterns in the
win pool and the place pool (being American it has an element of the show
pool included)

This system is involved, if you follow it to the letter, but again its
object is to identify those runners under bet in the place pool, relative to
their pool portion in the win pool.

Another variation, continuation, or some such of 'Place Power' -see Beat the
Racetrack and Betting at the Racetrack (Ziemba and Hausch) two books full of
algebra.

I don't want to seem to be disparaging about algebraic maths, finding a
winner is complicated enough, without finding a value for C, when A and B
are minute numbers, and those with another letter in them.

Ok, I'm disparaging, deriding, but only because I don't understand.

Cheers

Tony

From: Racing [mailto:racing-bounces at ausrace.com] On Behalf Of L.B.Loveday
Sent: Monday, April 16, 2018 8:57 PM
To: 'AusRace Racing Discussion List' <racing at ausrace.com
<mailto:racing at ausrace.com> >
Subject: Re: [AusRace] Place Power - a system

Tony,

I calculated Winx on final Vic TAB which I see you used, and got 2.37, so
that matched, but had differences with others (eg R2, Renewal I got 1.02).

Not to matter, I calculated 33,000 races (360,000 runners), minimum starters
8, no late scratchings, using Tatts Final Dividends, mainly to be 100% sure
that what I "knew" held up.

It did - the average ratios when Final Win Dividend:

<2.0, 2.12  

2.0 & <5.0 1.35

5.0 & <10.0 0.95

10.0 & <25.0 0.74

25.0 & <50.0 0.63

>=50.0 0.55

Cheers,

LBL

From: Racing <racing-bounces at ausrace.com <mailto:racing-bounces at ausrace.com>
> On Behalf Of Tony Moffat
Sent: Sunday, 15 April 2018 12:34 PM
To: racing at ausrace.com <mailto:racing at ausrace.com> 
Subject: [AusRace] Place Power - a system

There was a time when electronic devices, communication articles the signs
said, were banned on course. Computers, calculators, mobile phone bricks,
anything interpreted as a communication article, you had to take them off
course, you had to leave and take the devils tools with you. There were
signs about this, at the entrances - no phones basically

Place Power was a software package, chosen for installation on Canon
calculators, with programs for specific HP, TI and other brands of the time
- these were programmable calculators with minimal RAM and the program in
the ROM and so scientific. I had an upmarket Texas Instrument (I worked for
an affiliate of theirs for some years) which was hard on battery power but
had a proper sized keyboard. It was a brilliant piece of hardware for the
time, blue and black, with a large screen, and of a size that was a boast, a
display of knowledge that perhaps you didn't have but the device might. It
came with a holster, to complete the manliness aspect, or, a vest with a
sizeable elasticised pocket for it to be warmed and comforted by contact.
That vest was unlikely to be worn on course I suggest. There was a pen, a
plastic rod, that you used to select the keys, they had a moulded hole in
their centre to accept the point, this was to prevent moisture, sweat then,
from contacting the facia, where the keys were. They were sealed against
Resch Pilsner, I tested that aspect. No, I never used it in the bath, or
shower. It is military grade, volo 7, and meant for the Moon and Mars and
beyond, truly.

You purchased the calculator wherever you could get the best deal, then you
sent it to Place Power who loaded up the program, they said it was specific
to each brand, each model. It came back to you in the mail, wrapped in that
bubble wrap stuff, kids love it, and you went ahead with making your
fortune. My program had a lot of other data with it, statistics for
barriers, TAB numbers, parlay programs, all up programs, about 30 data cases
in all (34 actually). 

There is a downside to all this scientific magnificence, this being when
they purged the ROM to install their program, they wiped the surveying
program and the navigation program.

It was expensive, the program, the purchase of the calculator, the time
element. It can be torture, in the seconds before race starting time,
keying in the data elements, hitting RUN, and waiting for the BASIC program
to enliven the screen with the data you need, needed. The technique,
perhaps, was to enter all the data with 10 minutes to start time, run it,
then update only those runners which were sorted to the top of the list.

The day-ta or da-ta, it was spelt that way in the book, 230 pages of an
advertisement for the program, was the win and place dividends. The program
displayed these as % values of the whole dividend field, the book %, then
reduced this to a 100% market, for win divs. It did the same for place, the
300% market was reduced to 100%. The win dividend win % and the place
dividend % were compared with stored data and if there was a variation, more
money on the win side, or more money on the place side, or even over the
whole bet then the calculator did its thing and informed you and made a
betting suggestion, just off this. 

The runner(s) to be supported were those which had less % in the place line
than in the win line. The instruction was that more money had been bet, for
the win, than for the place, this shows confidence, this is information not
available to anybody (else - except system users).

The method, the rationale of all of this is investigated several times in
academic texts. The data revealed is used  by several authors, who take the
first findings forward in an endeavour to correctly isolate a runner with
the best credentials, off investment input. The writings are heavy on
algebra, mystic like, with assertions that .0062 is marvellous whilst .0053
is not, useless.

A starting point might be Peter Asch and Richard E Quandt 'Racetrack
Betting' or/and 'Market Efficiency in Racetrack Betting'. Google for more,
heaps, to be non-scientific about it at all. 

The endeavours of them all in attempting to straighten the line, the arrow
that points to a dividend, is appreciated. Them and their regression
analysis equations, god bless 'em.

The process in the program is described in detail in Asch and Quandt (1986)
page 117 onwards, although it, the process, is not given ownership to
anybody in the expansive book that comes with the program. It just is, with
flowery adjectives describing how it is good, better, best. The principle
was known well before the date of manufacture of the calculator. Perhaps it
was a parallel development, he asks mockingly.

When I returned to the program supplier, to have the navigation and
surveying programs re-installed several years later, he told me they had
sold two copies of Place Power, although copying, pirating was a common
activity back then, especially for HP model 41 schemes and programs. I still
have mine, giant robust device that it is, it feels like you could open
bottle tops with it, hammer nails, and a child has had it as a cot toy,
although this was not planned, honestly, it has no taste, no flavour. It has
a heft which is comforting, and it fires up instantaneously, still does.

If a runner had 14% of the win pool and 12%, or 11% of the place or some
figure less, this would ear mark itself as a runner to do further research,
especially if these differences occurred later in the betting, less than 100
seconds say. It is assumed that all the money is in the pool then, bettors
with known information, bettors with private information, all ups, and later
money, after 100 seconds, it is an assumption though, is from bettors,
little and large, private and corporate, betting into a niche now revealed.
Who knows this? 

The runner information, the dividend clauses, is similar to test done by
several persons in several way (days since last start, form, last start
finish position etc), and further, the runner to be supported must be 1.2
times less than the win dividend per centage holding, found by dividing the
place percent dividend holding into the win dividend place holding.

Starting with Randwick R1 yesterday - selections were scored from after the
race data, the final dividends.

The win dividend % holding was 10.1, the place dividend % holding was 8.5,
10.1/8.5 = 1.18 (it was 1.3 when the decision to select was made) result 2nd
$2.6.

Race 2 1.53 3rd $1.60

Race 3 1.6 2nd $2.3 - the system selected 1st,2nd and 3rd

Race 4  1.22 1st $2.4

Race 5 1st 1.7 $2.9

Race 6 1st 1.47 $1.8, the system selected 2nd also

Race 7 1st 2.36 $1.04 -this was Winx

Race 8 3rd 1.22 $1.7

Race 9 2nd 1.19 (it was 1.4 when selected) $3.7 -there was lots of
electronic action on this race, the get out stakes v1

Race 10 1st 1.41 $2.4 - there was lots of electronic action on this race,
the get out stakes v2

It selected a winner, a dividend, in every race bar the 1st at Caulfield -
it made no selection in this race.

I did 33 races, on a spreadsheet program (Smartbet v2.05), and it selected a
dividend in all of them. Too good to be true?

Summary: Compare the win dividend with the place dividend - there are
several ways.

If the win dividend appears oversubscribed, determine if the place dividend
is a value bet now, $W/$P and if the runner has other attributes (decided by
you) consider it for a bet. This last clause is necessary to reduce
qualifiers, there can be 3 or so, depending on the betting volatility, and
is suggested in the book with the program.

Winx was a selection. She had a dividend score of 2.36 (Win divided by
place) and other winning attributes.

Cheers

Tony

<http://www.avg.com/email-signature?utm_medium=email&utm_source=link&utm_cam
paign=sig-email&utm_content=emailclient> 

Virus-free.
<http://www.avg.com/email-signature?utm_medium=email&utm_source=link&utm_cam
paign=sig-email&utm_content=emailclient> www.avg.com 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://ausrace.com/pipermail/racing_ausrace.com/attachments/20180418/afc8690f/attachment.html>